Sudden death secondary to ventricular fibrillation (VF) remains a leading cause of mortality in the US. Therapy for VF has been largely ineffectual, principally because the underlying mechanisms for VF are not well understood and probing for potential mechanisms has been hindered by the inability to precisely modify specific ionic currents. To address these issues, we propose to develop a data-driven computer model of the electrical behavior of the canine ventricle. Specifically, we will: 1) Experimentally characterize IKr, ICa, IK1, INaCa, and the late sodium current INa in myocytes obtained from specific regions of the ventricles. These particular currents will be studied because they play a significant role in repolarization. They will be measured using action potentials recorded at rapid pacing rates as the command waveforms, to replicate current behavior during a tachyarrhythmia. 2) Develop deterministic Hodgkin-Huxley and Markov models for each ionic current for each anatomical region using the time series and steady state current data obtained under Specific Aim 1. Optimization routines will be used to determine unknown parameters in the models by comparing the model current to experimental data. 3) Incorporate the models of the individual currents into computer models of region-specific single canine ventricular myocytes. Models of left and right ventricular epicardial, midmyocardial and endocardial myocytes of basal and apical origin and of right and left ventricular Purkinje myocytes will be developed. 4) Incorporate the single cell models into a 3-D computer model of the canine ventricle using a modified version of the phase field method. The model will be written using a portable parallel version of the code and run on a parallel computer and multi-node clusters. Initially, the model will consist of the left ventricle, with epicardial, midmyocardial and endocardial layers. More detailed anatomical models subsequently will be constructed to include the His-Purkinje system and the right ventricle. 5) Use the 3-D model to test candidate hypotheses for the development of VF. The initial test will determine whether suppressing dynamic electrical heterogeneity prevents VF. The computer model of ventricular electrical function we propose will provide an invaluable tool for drug discovery and the evaluation of algorithms for anti-tachycardia and anti-fibrillatory pacing and defibrillation. As such, the model is expected to have a significant impact on the diagnosis and treatment of lethal heart rhythm disorders.